def merge_by_glycopeptide_sequence(matches):
groups = collectiontools.groupby(matches.values(), lambda x: x["Glycopeptide_identifier"], kind="sqlist")
merged_matches = sqlitedict.SqliteDict(matches.filename, tablename="matched_ions")
cntr = 0
for glycopeptide, matches in groups.items():
merged_matches[cntr] = merge_matches(matches)
cntr += 1
merged_matches.commit()
return merged_matches
def merge_ion_matches(matches):
groups = collectiontools.groupby(itertools.chain.from_iterable(matches),
lambda x: x["key"])
best_matches = []
fabs = math.fabs
for key, matched_key in groups.items():
best_match = matched_key[0]
best_ppm = fabs(best_match["ppm_error"])
ppm_to_scan_id = {best_match["scan_id"]: best_match["ppm_error"]}
for match in matched_key[1:]:
ppm_to_scan_id[match["scan_id"]] = match["ppm_error"]
if fabs(match["ppm_error"]) < best_ppm:
best_match = match
best_ppm = fabs(match["ppm_error"])
best_match["scan_map"] = ppm_to_scan_id
best_matches.append(best_match)
return best_matches